Hi Lorenzo,
I've taken the liberty of cc'ing the Maxima list
since this topic might be of general interest.
I wrote:
> > An idea here is to translate LBFGS (a Fortran code)
> > from Netlib.
I've successfully processed lbfgs.f with f2cl,
a Fortran->Lisp translator. I'll let you know
when I've verified the translated code is
working as intended.
You wrote:
> I wrote a simply script with maxima, which I attach
> (it was intended for my own personal use; it is not
> "professionally written" for sure. It mainly
> implements a few basic formulas in portfolio
> investment with very basic constraints, so that
> analytic formulas are available. It reads the return
> of a set of N assets [something you can sell/buy and
> which has a certain return \mu_i and a certain std
> \sigma_i] each containing Ndata elements. The code
> works the covariance matrix and a few other things
> and then straightforwardly implements some formulas
> minimizing the std for a chosen expected return mu of
> the portfolio).
Thanks for sending your script. I have an idea
about programming that could be useful here.
Maybe we can organize the results like this --
asset_analysis (mu=123.4, sigma=0.324, foo=1102, n=12)
-- instead of making mu, sigma, foo, etc global variables.
"asset_analysis" looks like a function name, but there
doesn't need to be a function by that name.
Incidentally this is the way R organizes results.
A linear regression is something like my.result <- lm (whatever),
then summary (my.result) shows a nicely formatted report,
and dput (my.result) shows every last datum in my.result.
This is especially convenient when my.result is big and messy.
Thanks again for your interest in Maxima. Although
you haven't much experience with Maxima, clearly you
understand the domain of econometrics. This expertise
is very valuable to Maxima.
regards,
Robert Dodier
____________________________________________________
Yahoo! Sports
Rekindle the Rivalries. Sign up for Fantasy Football
http://football.fantasysports.yahoo.com